Ann. Zool. Fennici 42: 433–447 ISSN 0003-455XHelsinki 29 August 2005 © Finnish Zoological and Botanical Publishing Board 2005
Effects of habitat fragmentation at different trophic levels in insect communities
Saskya van Nouhuys
Metapopulation Research Group, Department of Biological and Environmental Sciences, Division of Population Biology, P.O. Box 65 (Viikinkaari 1), FI-00014 University of Helsinki, Finland; and Department of Ecology and Evolutionary Biology, Corson Hall, Cornell University, Ithaca, NY 14853, USA (e-mail address: [email protected])
Received 21 Jan. 2005, revised version received 19 Apr. 2005, accepted 20 Apr. 2005
van Nouhuys, S. 2005: Effects of habitat fragmentation at different trophic levels in insect commu-nities. — Ann. Zool. Fennici 42: 433–447.
Species experience landscapes differently depending on their needs and behaviors, and on their trophic level. We expect species at high trophic levels in a community to be more sensitive to habitat fragmentation than species at lower trophic levels. But this depends on attributes such as resource breadth, dispersiveness, reproductive rate, and longevity, which may not be related to trophic level. I address the association of frag-mentation with trophic level using a literature review of 31 studies of herbivores and their natural enemies, and a case study of the parasitoids associated with the Glanville fritillary butterfly. Measures of species richness or total parasitism in an entire insect community provide the strongest support for the idea that negative effects of frag-mentation amplify at higher trophic levels. Generally though, there is great variation among studies, due to variation among species, as well as in designs of both experi-mental and observational studies.
Introduction
As a landscape becomes fragmented, changes occur in the insect (meta)community inhabiting it. Understanding the mechanisms behind these changes is a task for contemporary community ecology. We expect different ecological and evo-lutionary responses to landscape change among species based on the breadth of their resource needs and their behavior. At one extreme individ-uals may be dispersive and long-lived, moving at such a large scale that they do not perceive the landscape as patchy. Similarly, a species that uses a broad range of resources may also experi-ence the landscape as continuous. At the other extreme are sedentary species made up of indi-
viduals that usually spend their entire life within a single habitat patch. Correspondingly, if the resource requirements of a species are extremely narrow, it will perceive the landscape as highly fragmented because only a small fraction of the total area is useful (MacArthur & Levins 1964).
We also expect different responses to habitat fragmentation depending on the trophic level of a species. This is because food/prey/host species are limited to a subset of the suitable locations in the landscape, and because local population dynamics of species at the lower trophic level make them an unstable resource for the depend-ent species (Holt 2002, van Nouhuys & Hanski 2005). In some ways a fragmented habitat is simply a type of low quality habitat (summa-
434 van Nouhuys • ANN. ZOOL. FENNICI Vol. 42
rized by Morin 1999) in which food chain length decreases with habitat quality. Within a single fragment species persistence decreases with fragment size. At the landscape scale, small total fragment area and long distance between frag-ments can decrease regional persistence. Where primary productivity is low there is not reliable food to support persistence of high trophic level species. In this context the prediction, and gen-eral observation, is that high trophic level species are absent from fragmented habitats (see summa-ries in Pimm 1982, Mikkelson 1993, Didham et al. 1996, Tscharntke & Kruess 1999, Holt 2002, Henle et al. 2004). While this is probably true, it is certainly not always the case.
As long as habitat fragmentation is not too extreme, fragmentation can promote regional coexistence of predators and prey (Hassell 2000, Holt 2002), and its effect can differ greatly among species. Attributes of species that affect their sensitivity to habitat fragmentation, such as dispersal behavior, rarity, reproductive poten-tial, habitat specificity, competitive ability, and biogeography (Henle et al. 2004), do not nec-essarily correspond to trophic level. In fact, it may be that qualities that are advantageous at high trophic levels are also advantageous in fragmented landscapes. For example hyperpara-sitoids (4th trophic level or higher) tend to be ectoparasitic rather than endoparasitic (Brodeur 2000). While we know too little about the host ranges of hyperparasitoids to say how host spe-cific they actually are, we do know that ectopara-sitoids generally have much broader host ranges than endoparasitoids (Shaw 1994). Additionally many hyperparasitoids are facultative primary parasitoids. Both of these characteristics, broad host range and multitrophic level feeding, would decrease the negative effects of fragmentation by allowing the hyperparasitoid to make do with what is available in a given habitat fragment.
Because higher trophic levels experience a more fragmented habitat we expect them to suffer more from fragmentation than the spe-cies below them. On the other hand, the effects of fragmentation depend on attributes of the insect that may be unrelated to trophic level, or even predispose higher trophic level species to perform well in fragmented landscapes. With no clear conceptual resolution I will now approach
this question using empirical studies from the literature, and a case study from my own work on the parasitoid community associated with the Glanville fritillary butterfly in Finland.
Literature review
This is a literature survey of empirical research papers that contain data about the impact of hab-itat fragmentation on herbivores and their para-sitoids or predators. The papers surveyed were collected, in part, by searching ecological, con-servation, and agricultural electronic databases using the key word “parasitoid” in combination with “fragmentation,” “landscape,” or “disper-sal.” In all there are 31 papers about 23 different research systems (Table 1). There are three types of studies. The first are 11 observational studies of naturally fragmented landscapes. An example of such a study is the work by Antolin and Strong (1987) showing that both the leaf hopper Proke-lesia marginata and its egg parasitoid Anagrus delicatus routinely disperse up to one km to off-shore islands in their fragmented salt marsh hab-itat. A contrasting study is that of Eber (2001), in which the agromyzid leaf miner Phytomyza ilicis and its parasitoids move very little among patchily distributed holly trees. The second cat-egory of data is from eight observational stud-ies of anthropogenically fragmented landscapes. These studies include the effects of old growth forest fragmentation on the insects associated with bracket fungus (Komonen et al. 2000), and the insect communities associated with isolated nettle stands in an agricultural landscape (Zabel & Tscharntke 1998). Finally there are 11 manip-ulated fragmentation experiments, such as the study by Kruess and Tscharntke (2000b) of bush vetch (Vicia sepium) colonization by seed pod feeding insects and their parasitoids between 100 and 500 m from established vetch patches.
This distinction between natural and anthro-pogenic fragmentation is often subjective. For example, I categorized the holly bush study by Eber (2001) as naturally fragmented even though the distribution of trees is likely influenced by humans, and the study of Silene latifolia (J. A. Elzinga et al. unpubl. data) as anthropogenically fragmented because it grows in agriculturally
ANN. ZOOL. FENNICI Vol. 42 • Fragmentation and trophic level 435T
able
1. S
umm
ary
of s
tudi
es u
sed
in li
tera
ture
rev
iew
of t
he e
ffect
s of
hab
itat f
ragm
enta
tion
on in
sect
s an
d th
eir
para
sito
ids
and
pred
ator
s.
Cit.
no.
1 A
utho
rs
Typ
e of
com
mun
ity
Spa
tial a
nd te
mpo
ral s
cale
and
E
ffect
s of
frag
men
tatio
n3
ty
pe2
of s
tudy
01
Am
aras
ekar
e S
age
scru
b, Is
omer
is
Exp
erim
ent:
repl
icat
e cl
uste
rs o
f Is
olat
ion:
no
effe
ct o
n 1
H; –
effe
ct
2000
a, 2
000b
patc
hes
sepa
rate
d by
20
and
40 m
. on
2 P
col
oniz
atio
n.
1.5
year
s (=
5 H
gen
erat
ions
&
14
–16
P g
ener
atio
ns).
02
A
ntol
in &
C
od g
rass
, Spa
rtin
a al
tern
ifolia
N
atur
al: S
ticky
trap
s on
isle
ts
Isol
atio
n: n
o ef
fect
on
abun
danc
e of
S
tron
g 19
87
an
d m
ainl
and
(abo
ut 1
km
1
H, +
effe
ct o
n 1
P.
ap
art)
. 1 s
easo
n.
03
Bra
schl
er e
t al.
Cal
care
ous
gras
slan
d E
xper
imen
t: re
plic
ated
pat
ches
P
atch
siz
e: +
for
H (
mul
tiple
aph
id
2003
(0.5
to 4
.5 m
2 ) e
ach
5 m
apa
rt,
spec
ies)
den
sity
, no
effe
ct o
n H
an
d un
frag
men
ted
cont
rol p
lots
. sp
ecie
s ric
hnes
s or
div
ersi
ty, n
o
3 ye
ars.
ef
fect
on
P.
04
Cap
pucc
ino
et
Bal
sam
fir,
Abi
es b
alsa
mea
A
nthr
opog
enic
: unf
ragm
ente
d Is
olat
ion:
no
effe
ct fo
r 1
H (
spru
ce
al. 1
998
fore
st
fore
st, “
habi
tat i
slan
ds”
(0.2
5 to
bu
dwor
m);
– fo
r 1
para
sito
id; +
for
2.
0 ha
) an
d “t
rue
isla
nds”
. 2
4 pa
rasi
toid
s on
hab
itat i
slan
ds b
ut
grow
ing
seas
ons.
no
t tru
e is
land
s.05
C
roni
n &
C
od g
rass
, Spa
rtin
a al
tern
ifolia
E
xper
imen
t: re
plic
ate
patc
hes
Isol
atio
n (p
er s
e): n
o ef
fect
for
1 P
H
ayne
s 20
04
(0
.7 ¥
0.9
m),
in c
lust
ers
(5 m
or
1 H
. Mat
rix: H
& P
mov
e m
ore
sp
acin
g) 3
0 m
apa
rt, i
n br
ome
or
in b
rom
e. L
ocal
ext
inct
ion
of H
m
udfla
t mat
rix. 1
sea
son.
hi
gher
in b
rom
e (s
tron
ger
for
P).
06
Cro
nin
2003
C
od g
rass
, Spa
rtin
a al
tern
ifolia
N
atur
al: 2
5 to
147
patc
hes
(0.1
Is
olat
ion:
– fo
r 1
H a
nd 1
P
to 1
26.7
m2 ;
1 to
46
m a
part
) in
(s
tron
ger
for
P).
Pat
ch s
ize
and
65
ha.
3.5
yea
rs (
7 H
de
nsity
: – fo
r 1
H a
nd 1
P.
ge
nera
tions
). E
xtin
ctio
n an
d M
atrix
: mud
flat –
for
1 H
and
+ 1
P.
co
loni
zatio
n dy
nam
ics.
07
C
roni
n et
al.
Pin
e fo
rest
Pin
us ta
eda,
P.
Mar
k re
capt
ure
expe
rimen
t: D
ista
nce
mov
ed: H
less
than
Pr
20
00
echi
nata
tr
aps
plac
ed 0
.1 to
5 k
m fr
om
(95%
of H
mov
ed le
ss th
an 2
.3 k
m,
or
igin
. 1 s
ampl
ing
perio
d.
95%
of P
r m
oved
less
than
5.1
km
).08
C
roni
n et
al.
Cod
gra
ss, S
part
ina
alte
rnifo
lia
Nat
ural
: 147
pat
ches
(0.
1 to
Is
olat
ion:
– fo
r 1
H; n
o ef
fect
on
Pr.
20
04, C
roni
n (s
pide
rs &
inse
cts)
12
6.7
m2 ;
1 to
46
m a
part
) in
65
Pat
ch s
ize:
– fo
r 1
H; +
for
Pr.
20
03
ha
. 1.5
yea
rs (
3 H
gen
erat
ions
).
Mat
rix: m
udfla
t – fo
r 1
H; n
o ef
fect
on P
r.09
D
ubbe
rt e
t al.
Gra
ss, C
alam
agro
stis
epi
geio
s N
atur
al: 2
5 pa
tche
s (1
6 to
500
0 Is
olat
ion:
no
effe
ct o
n 4H
; – e
ffect
19
98
m
2 ; 0
to 1
30 m
apa
rt; 0
.5 to
95
on r
are
2H; –
effe
ct 1
out
of 1
8 P
.
shoo
ts p
er m
2 ). 1
–2 s
easo
ns.
Pat
ch s
ize:
no
effe
ct o
n 4H
or
18 P
.
P
lant
den
sity
: – e
ffect
on
4H; –
effe
ct s
ome
P (
thro
ugh
effe
ct o
n H
).
cont
inue
d
436 van Nouhuys • ANN. ZOOL. FENNICI Vol. 42T
able
1. C
ontin
ued.
Cit.
no.
1 A
utho
rs
Typ
e of
com
mun
ity
Spa
tial a
nd te
mpo
ral s
cale
and
E
ffect
s of
frag
men
tatio
n3
ty
pe2
of s
tudy
10a
Ebe
r 20
01
Mea
dow
, cre
epin
g th
istle
, N
atur
al: 3
5 km
2 . L
ong
term
Is
olat
ion:
no
effe
ct o
n 1
H a
nd
C
risiu
m a
rven
se
stud
y.
seve
ral P
. Pat
ch s
ize:
– e
ffect
on
1
H
and
sev
eral
P.
10b
Ebe
r 20
01
Hol
ly, I
lex
aqui
foliu
m
Nat
ural
: 15
patc
hes
in a
1.5
km
2 Is
olat
ion:
no
effe
ct o
n 1
H a
nd
area
. Lon
g te
rm s
tudy
. se
vera
l P. P
atch
siz
e: n
o ef
fect
on
1
H
and
sev
eral
P.
11a
J. A
. Elz
inga
et
Whi
te c
ampi
on, S
ilene
bic
ruris
N
atur
al: 8
5 pa
tche
s (1
to 1
3000
Is
olat
ion:
no
effe
ct o
n H
; – e
ffect
on
al
. unp
ubl.
data
plan
ts; a
t lea
st 1
00 m
par
t), 1
00
1 of
3 P
. Pat
ch s
ize:
no
effe
ct o
n H
;
km a
long
riv
er. 3
sea
sons
. –
effe
ct o
n al
l P.
11b
J. A
. Elz
inga
et
Whi
te c
ampi
on, S
ilene
bic
ruris
E
xper
imen
t: re
plic
ated
pat
ches
Is
olat
ion:
no
effe
ct o
n H
; – e
ffect
on
al
. unp
ubl.
data
125
to 2
000
m a
part
. 2 s
easo
ns.
1 of
3 P
.12
F
aeth
&
Oak
land
s, Q
uerc
us s
p.
Exp
erim
ent:
3 is
olat
ed tr
ees
Isol
atio
n: –
for
colo
niza
tion
by 2
H,
S
imbe
rloff
1981
tran
spla
nted
165
m fr
om fo
rest
. +
for
1H, n
o ef
fect
on
3 H
2
seas
ons.
ab
unda
nce;
– fo
r ra
te o
f par
asiti
sm,
no e
ffect
for
rate
of P
reda
tion.
13
Fer
rari
et a
l. W
ild r
oses
, Ros
a ca
nina
, R.
Nat
ural
: 31
patc
hes.
1 s
easo
n.
Isol
atio
n: n
o ef
fect
2 H
; – 9
P (
tota
l
1997
co
rym
bife
ra
pa
rasi
tism
). P
atch
siz
e: –
2 H
; no
effe
ct 9
P (
tota
l par
asiti
sm).
14
Gol
den
& C
rist
Gol
denr
od, S
alid
ago
cana
dens
is
Exp
erim
enta
l: re
plic
ate
plot
s F
ragm
enta
tion:
No
effe
ct o
n H
, P
1999
mad
e of
rep
licat
ed p
atch
es (
1, 2
or
Pr
abun
danc
e (m
ultip
le s
peci
es);
&
3 m
2 ). 1
sea
son.
sm
all –
effe
ct o
n H
div
ersi
ty; +
effe
ct o
n 2H
spe
cies
; – r
are
spec
ies.
15
Har
rison
&
Mea
dow
, rag
wor
t, S
enec
io
Nat
ural
: 60
patc
hes
(12
to 1
04 Is
olat
ion:
no
effe
ct o
n 10
H a
nd 1
P.
T
hom
as 1
991
jaco
baea
m
2 ; 1
0–10
0 m
apa
rt)
in a
1.3
D
istu
rban
ce: +
and
– e
ffect
s on
H,
km
2 ar
ea. 1
sea
son.
bu
t not
P.
16
Kar
eiva
198
7 M
eado
w, g
olde
nrod
, Sol
idag
o sp
. E
xper
imen
tal:
cont
inuo
us v
s. 1
-m2
Fra
gmen
tatio
n: +
for
1 H
and
1 P
r
bloc
ks 1
m a
part
. 4 s
easo
ns.
abun
danc
e bu
t not
col
oniz
atio
n.17
K
omon
en e
t al.
For
est b
rack
et fu
ngus
A
nthr
opog
enic
: 8 u
nfra
gmen
ted
Isol
atio
n: –
for
F a
nd P
; P a
bsen
t
2000
fore
sts,
4 y
oung
isol
ated
fr
om o
ld fr
agm
ent.
fr
agm
ents
, and
8 o
ld is
olat
ed
frag
men
ts. 1
sam
plin
g pe
riod.
18
Kru
ess
&
Mea
dow
, red
clo
ver,
Trif
oliu
m
Exp
erim
enta
l: 18
pat
ches
(1.
2 m
2 ;
Isol
atio
n: –
for
abun
danc
e &
T
scha
rntk
e 19
94
prat
ense
se
para
ted
by 1
00 to
500
m).
1
dive
rsity
of 8
H (
stro
nger
for
rare
se
ason
. sp
ecie
s); –
for
12 P
div
ersi
ty a
nd %
para
sitis
m (
stro
nger
for
P th
an fo
r
H
).
19a
Kru
ess
&
Mea
dow
, bus
h ve
tch,
Vic
ia
Ant
hrop
ogen
ic: 1
8 m
eado
ws
Pat
ch a
rea:
– s
peci
es d
iver
sity
of 4
T
scha
rntk
e se
pium
(3
00 m
to 7
0 ha
). 1
sea
son.
H
and
10
P (
stro
nger
for
P th
an H
);
2000
a, 2
000b
–
abun
danc
e of
1 H
; – r
ate
of
pa
rasi
tism
; – r
are
spec
ies.
19b
Kru
ess
&
Mea
dow
, bus
h ve
tch,
Vic
ia
Exp
erim
enta
l: 16
pat
ches
(12
Is
olat
ion:
– s
peci
es d
iver
sity
4 H
T
scha
rntk
e se
pium
pl
ants
eac
h) is
olat
ed b
y 10
0 to
an
d 2
P; –
abu
ndan
ce o
f 2 o
ut o
f 4
2000
b
500
m, e
ach
paire
d w
ith
H a
nd 2
out
of 2
P.
un
isol
ated
pla
nts.
1 s
easo
n.
No
effe
ct o
n to
tal p
aras
itism
, but
–
fo
r 1
P; –
rar
e sp
ecie
s.20
K
rues
s &
M
eado
w, r
ed c
love
r, T
rifol
ium
A
nthr
opog
enic
: 18
mea
dow
s Is
olat
ion:
no
effe
ct. P
atch
siz
e: –
T
scha
rntk
e pr
aten
se
(300
m to
70
ha),
isol
ated
by
0 fo
r di
vers
ity o
f 8 H
and
15
P
2000
a
to 5
00 m
eter
s. 1
sea
son.
(s
tron
ger
for
P);
– %
par
asiti
sm.
21
Kru
ess
2003
M
eado
w, c
reep
ing
this
tle,
Ant
hrop
ogen
ic: 1
5 pa
tche
s (a
t Is
olat
ion:
– fo
r so
me
H, –
9 P
.
C
risiu
m a
rven
se
leas
t 24
m a
part
), in
cro
p,
Ric
hnes
s an
d ab
unda
nce
and
%
mar
gin
and
fallo
w h
abita
t typ
es.
para
sitis
m. M
atrix
: no
effe
ct o
n 9
1
seas
on.
H; e
ffect
on
6P r
ichn
ess
and
abun
danc
e %
par
asiti
sm.
22
Rol
and
& T
aylo
r A
spen
fore
st, P
opul
us te
mul
oide
s A
nthr
opog
enic
: Deg
ree
of fo
rest
F
ragm
enta
tion:
– fo
r 1
H a
nd 3
P, +
19
97
fr
agm
enta
tion
estim
ated
aro
und
for
1 P
(m
easu
red
as %
par
asiti
sm).
po
ints
on
a 42
0 m
2 gr
id. 1
seas
on.
23
Seg
arra
- T
ropi
cal l
egum
e, C
rota
laria
N
atur
al: 6
2 pa
tche
s (2
0 to
200
Is
olat
ion:
not
mea
sure
d. P
atch
siz
e:
Car
mon
a &
pa
llida
pl
ants
, at l
east
100
0 m
apa
rt),
–
effe
ct fo
r 1
H a
nd 1
P (
stro
nger
B
arbo
sa 1
992
ea
ch s
ampl
ed a
t lea
st o
nce
over
fo
r P
than
H).
4
year
s.
24
Tsc
harn
tke
&
Ree
d, P
hrag
mite
s au
stra
lis
Ant
hrop
ogen
ic: 2
8 ha
bita
t P
atch
siz
e: –
for
pres
ence
of 2
H
Kru
ess
1999
;
frag
men
ts (
rang
e fr
om a
bout
an
d 8
P. P
atch
age
: – %
par
asiti
sm
Ath
en &
100
to 1
500
m2 ;
3 to
10
year
s of
2 H
.
Tsc
harn
tke
1999
old)
. 1 s
ampl
ing
perio
d.
25
Tsc
harn
tke
et a
l. M
eado
w, t
rap
nest
ing
bees
and
E
xper
imen
tal:
42 p
atch
es (
each
Is
olat
ion:
– d
iver
sity
of B
but
not
19
98
thei
r pa
rasi
toid
s m
ade
of 8
pot
s of
pla
nts
and
a to
tal a
bund
ance
; – d
iver
sity
of P
be
e tr
ap)
0 to
120
0 m
from
an
d P
r, a
nd %
par
asiti
sm a
nd
sour
ces.
1 s
easo
n.
pred
atio
n.26
va
n de
r M
eijd
en &
D
une,
rag
wor
t, S
enec
io ja
coba
ea
Nat
ural
: sev
eral
100
pat
ches
Is
olat
ion
and
patc
h si
ze: –
for
1 H
va
n de
r V
een-
van
Wijk
sepa
rate
d by
abo
ut 1
00 to
500
m.
and
1 P
(st
rong
er fo
r P
).
1997
Long
term
stu
dy.
27
van
Nou
huys
&
Mea
dow
, Ver
onic
a sp
icat
a &
A
nthr
opog
enic
: 400
0 m
eado
ws
Isol
atio
n an
d pa
tch
size
: – fo
r 1
H
Han
ski 1
999,
P
lant
ago
lanc
eola
ta
(300
–500
occ
upie
d by
H)
(100
co
loni
zatio
n an
d pe
rsis
tenc
e; –
for
20
02
m
2 to
10
ha; 1
00 to
abo
ut 5
000
1 P
col
oniz
atio
n an
d pe
rsis
tenc
e
m a
part
). L
ong
term
stu
dy.
(str
onge
r fo
r H
); n
o ef
fect
on
1 P
.
cont
inue
d
ANN. ZOOL. FENNICI Vol. 42 • Fragmentation and trophic level 437
Tab
le 1
. Con
tinue
d.
Cit.
no.
1 A
utho
rs
Typ
e of
com
mun
ity
Spa
tial a
nd te
mpo
ral s
cale
and
E
ffect
s of
frag
men
tatio
n3
ty
pe2
of s
tudy
10a
Ebe
r 20
01
Mea
dow
, cre
epin
g th
istle
, N
atur
al: 3
5 km
2 . L
ong
term
Is
olat
ion:
no
effe
ct o
n 1
H a
nd
C
risiu
m a
rven
se
stud
y.
seve
ral P
. Pat
ch s
ize:
– e
ffect
on
1
H
and
sev
eral
P.
10b
Ebe
r 20
01
Hol
ly, I
lex
aqui
foliu
m
Nat
ural
: 15
patc
hes
in a
1.5
km
2 Is
olat
ion:
no
effe
ct o
n 1
H a
nd
area
. Lon
g te
rm s
tudy
. se
vera
l P. P
atch
siz
e: n
o ef
fect
on
1
H
and
sev
eral
P.
11a
J. A
. Elz
inga
et
Whi
te c
ampi
on, S
ilene
bic
ruris
N
atur
al: 8
5 pa
tche
s (1
to 1
3000
Is
olat
ion:
no
effe
ct o
n H
; – e
ffect
on
al
. unp
ubl.
data
plan
ts; a
t lea
st 1
00 m
par
t), 1
00
1 of
3 P
. Pat
ch s
ize:
no
effe
ct o
n H
;
km a
long
riv
er. 3
sea
sons
. –
effe
ct o
n al
l P.
11b
J. A
. Elz
inga
et
Whi
te c
ampi
on, S
ilene
bic
ruris
E
xper
imen
t: re
plic
ated
pat
ches
Is
olat
ion:
no
effe
ct o
n H
; – e
ffect
on
al
. unp
ubl.
data
125
to 2
000
m a
part
. 2 s
easo
ns.
1 of
3 P
.12
F
aeth
&
Oak
land
s, Q
uerc
us s
p.
Exp
erim
ent:
3 is
olat
ed tr
ees
Isol
atio
n: –
for
colo
niza
tion
by 2
H,
S
imbe
rloff
1981
tran
spla
nted
165
m fr
om fo
rest
. +
for
1H, n
o ef
fect
on
3 H
2
seas
ons.
ab
unda
nce;
– fo
r ra
te o
f par
asiti
sm,
no e
ffect
for
rate
of P
reda
tion.
13
Fer
rari
et a
l. W
ild r
oses
, Ros
a ca
nina
, R.
Nat
ural
: 31
patc
hes.
1 s
easo
n.
Isol
atio
n: n
o ef
fect
2 H
; – 9
P (
tota
l
1997
co
rym
bife
ra
pa
rasi
tism
). P
atch
siz
e: –
2 H
; no
effe
ct 9
P (
tota
l par
asiti
sm).
14
Gol
den
& C
rist
Gol
denr
od, S
alid
ago
cana
dens
is
Exp
erim
enta
l: re
plic
ate
plot
s F
ragm
enta
tion:
No
effe
ct o
n H
, P
1999
mad
e of
rep
licat
ed p
atch
es (
1, 2
or
Pr
abun
danc
e (m
ultip
le s
peci
es);
&
3 m
2 ). 1
sea
son.
sm
all –
effe
ct o
n H
div
ersi
ty; +
effe
ct o
n 2H
spe
cies
; – r
are
spec
ies.
15
Har
rison
&
Mea
dow
, rag
wor
t, S
enec
io
Nat
ural
: 60
patc
hes
(12
to 1
04 Is
olat
ion:
no
effe
ct o
n 10
H a
nd 1
P.
T
hom
as 1
991
jaco
baea
m
2 ; 1
0–10
0 m
apa
rt)
in a
1.3
D
istu
rban
ce: +
and
– e
ffect
s on
H,
km
2 ar
ea. 1
sea
son.
bu
t not
P.
16
Kar
eiva
198
7 M
eado
w, g
olde
nrod
, Sol
idag
o sp
. E
xper
imen
tal:
cont
inuo
us v
s. 1
-m2
Fra
gmen
tatio
n: +
for
1 H
and
1 P
r
bloc
ks 1
m a
part
. 4 s
easo
ns.
abun
danc
e bu
t not
col
oniz
atio
n.17
K
omon
en e
t al.
For
est b
rack
et fu
ngus
A
nthr
opog
enic
: 8 u
nfra
gmen
ted
Isol
atio
n: –
for
F a
nd P
; P a
bsen
t
2000
fore
sts,
4 y
oung
isol
ated
fr
om o
ld fr
agm
ent.
fr
agm
ents
, and
8 o
ld is
olat
ed
frag
men
ts. 1
sam
plin
g pe
riod.
18
Kru
ess
&
Mea
dow
, red
clo
ver,
Trif
oliu
m
Exp
erim
enta
l: 18
pat
ches
(1.
2 m
2 ;
Isol
atio
n: –
for
abun
danc
e &
T
scha
rntk
e 19
94
prat
ense
se
para
ted
by 1
00 to
500
m).
1
dive
rsity
of 8
H (
stro
nger
for
rare
se
ason
. sp
ecie
s); –
for
12 P
div
ersi
ty a
nd %
para
sitis
m (
stro
nger
for
P th
an fo
r
H
).
19a
Kru
ess
&
Mea
dow
, bus
h ve
tch,
Vic
ia
Ant
hrop
ogen
ic: 1
8 m
eado
ws
Pat
ch a
rea:
– s
peci
es d
iver
sity
of 4
T
scha
rntk
e se
pium
(3
00 m
to 7
0 ha
). 1
sea
son.
H
and
10
P (
stro
nger
for
P th
an H
);
2000
a, 2
000b
–
abun
danc
e of
1 H
; – r
ate
of
pa
rasi
tism
; – r
are
spec
ies.
19b
Kru
ess
&
Mea
dow
, bus
h ve
tch,
Vic
ia
Exp
erim
enta
l: 16
pat
ches
(12
Is
olat
ion:
– s
peci
es d
iver
sity
4 H
T
scha
rntk
e se
pium
pl
ants
eac
h) is
olat
ed b
y 10
0 to
an
d 2
P; –
abu
ndan
ce o
f 2 o
ut o
f 4
2000
b
500
m, e
ach
paire
d w
ith
H a
nd 2
out
of 2
P.
un
isol
ated
pla
nts.
1 s
easo
n.
No
effe
ct o
n to
tal p
aras
itism
, but
–
fo
r 1
P; –
rar
e sp
ecie
s.20
K
rues
s &
M
eado
w, r
ed c
love
r, T
rifol
ium
A
nthr
opog
enic
: 18
mea
dow
s Is
olat
ion:
no
effe
ct. P
atch
siz
e: –
T
scha
rntk
e pr
aten
se
(300
m to
70
ha),
isol
ated
by
0 fo
r di
vers
ity o
f 8 H
and
15
P
2000
a
to 5
00 m
eter
s. 1
sea
son.
(s
tron
ger
for
P);
– %
par
asiti
sm.
21
Kru
ess
2003
M
eado
w, c
reep
ing
this
tle,
Ant
hrop
ogen
ic: 1
5 pa
tche
s (a
t Is
olat
ion:
– fo
r so
me
H, –
9 P
.
C
risiu
m a
rven
se
leas
t 24
m a
part
), in
cro
p,
Ric
hnes
s an
d ab
unda
nce
and
%
mar
gin
and
fallo
w h
abita
t typ
es.
para
sitis
m. M
atrix
: no
effe
ct o
n 9
1
seas
on.
H; e
ffect
on
6P r
ichn
ess
and
abun
danc
e %
par
asiti
sm.
22
Rol
and
& T
aylo
r A
spen
fore
st, P
opul
us te
mul
oide
s A
nthr
opog
enic
: Deg
ree
of fo
rest
F
ragm
enta
tion:
– fo
r 1
H a
nd 3
P, +
19
97
fr
agm
enta
tion
estim
ated
aro
und
for
1 P
(m
easu
red
as %
par
asiti
sm).
po
ints
on
a 42
0 m
2 gr
id. 1
seas
on.
23
Seg
arra
- T
ropi
cal l
egum
e, C
rota
laria
N
atur
al: 6
2 pa
tche
s (2
0 to
200
Is
olat
ion:
not
mea
sure
d. P
atch
siz
e:
Car
mon
a &
pa
llida
pl
ants
, at l
east
100
0 m
apa
rt),
–
effe
ct fo
r 1
H a
nd 1
P (
stro
nger
B
arbo
sa 1
992
ea
ch s
ampl
ed a
t lea
st o
nce
over
fo
r P
than
H).
4
year
s.
24
Tsc
harn
tke
&
Ree
d, P
hrag
mite
s au
stra
lis
Ant
hrop
ogen
ic: 2
8 ha
bita
t P
atch
siz
e: –
for
pres
ence
of 2
H
Kru
ess
1999
;
frag
men
ts (
rang
e fr
om a
bout
an
d 8
P. P
atch
age
: – %
par
asiti
sm
Ath
en &
100
to 1
500
m2 ;
3 to
10
year
s of
2 H
.
Tsc
harn
tke
1999
old)
. 1 s
ampl
ing
perio
d.
25
Tsc
harn
tke
et a
l. M
eado
w, t
rap
nest
ing
bees
and
E
xper
imen
tal:
42 p
atch
es (
each
Is
olat
ion:
– d
iver
sity
of B
but
not
19
98
thei
r pa
rasi
toid
s m
ade
of 8
pot
s of
pla
nts
and
a to
tal a
bund
ance
; – d
iver
sity
of P
be
e tr
ap)
0 to
120
0 m
from
an
d P
r, a
nd %
par
asiti
sm a
nd
sour
ces.
1 s
easo
n.
pred
atio
n.26
va
n de
r M
eijd
en &
D
une,
rag
wor
t, S
enec
io ja
coba
ea
Nat
ural
: sev
eral
100
pat
ches
Is
olat
ion
and
patc
h si
ze: –
for
1 H
va
n de
r V
een-
van
Wijk
sepa
rate
d by
abo
ut 1
00 to
500
m.
and
1 P
(st
rong
er fo
r P
).
1997
Long
term
stu
dy.
27
van
Nou
huys
&
Mea
dow
, Ver
onic
a sp
icat
a &
A
nthr
opog
enic
: 400
0 m
eado
ws
Isol
atio
n an
d pa
tch
size
: – fo
r 1
H
Han
ski 1
999,
P
lant
ago
lanc
eola
ta
(300
–500
occ
upie
d by
H)
(100
co
loni
zatio
n an
d pe
rsis
tenc
e; –
for
20
02
m
2 to
10
ha; 1
00 to
abo
ut 5
000
1 P
col
oniz
atio
n an
d pe
rsis
tenc
e
m a
part
). L
ong
term
stu
dy.
(str
onge
r fo
r H
); n
o ef
fect
on
1 P
.
cont
inue
d
438 van Nouhuys • ANN. ZOOL. FENNICI Vol. 42
disturbed areas, even though it may have a natu-rally fragmented distribution as well. I attempt to distinguish between the two types of studies in order to separate insect communities that have a long evolutionary history in a habitat that we consider fragmented from those that have had a relatively short evolutionary history in a land-scape, and whose continued persistence in that environment is uncertain.
The data for these studies were generally gathered in three ways. Most were surveys of individual species pairs or communities of inter-acting species. Some of these surveys were con-ducted over several years in the same locations (van der Meijden & van der Veen-van Wijk 1997, van Nouhuys & Hanski 1999, 2002, Eber 2001), over multiple insect generations (Amar-asekare 2000b, Cronin 2004), or during one visit or over a season (Antolin & Strong 1987, Zabel & Tscharntke 1998, Kruess 2003). Herbivores were sampled or counted in place. Parasitoids were generally evaluated by the rate of parasit-ism of a sample of the host species collected from the research site (a sometimes misleading method of evaluation of parasitoids, see Askew & Shaw 1989). Several studies of the effects of fragmentation on specific taxa were conducted by trapping target species (Antolin & Strong 1987), and by mark-release-recapture (Cronin & Haynes 2004).
Fragmentation was evaluated in several dif-ferent ways as well. These include the size of fragments, connectivity among them, and their age and quality. Fragment size varied from 0.1 m2 tufts of grass (Cronin 2003, Cronin et al. 2004) to 70 ha fields (Kruess & Tscharntke 2000a, 2000b). The distances between fragments varied from one meter in the study of aphids and coccinelid bee-tles on goldenrod (Salidago sp.) (Kareiva 1987) up to several km in the study of the parasitoids of the Glanville fritillary (van Nouhuys & Hanski 1999, 2002). Connectivity was presented as the simple distance between patches or as a measure that one way or another took into account differ-ing contributions of patches based on size and distance (for a review and discussion of measures of connectivity see Moilanen & Nieminen 2004). Alternatively, a landscape was simply character-ized as fragmented or continuous without cal-culating connectivity among individual patches T
able
1. C
ontin
ued.
Cit.
no.
1 A
utho
rs
Typ
e of
com
mun
ity
Spa
tial a
nd te
mpo
ral s
cale
and
E
ffect
s of
frag
men
tatio
n3
ty
pe2
of s
tudy
28
Zab
el &
M
eado
w, s
tingi
ng n
ettle
, Urt
ica
Ant
hrop
ogen
ic: 3
2 pa
tche
s (2
to
Pat
ch s
ize:
– a
bund
ance
and
ric
hnes
s
Tsc
harn
tke
1998
di
oica
10
00 m
2 ), 2
1 is
olat
ed b
y 75
to
of H
; no
effe
ct o
n ric
hnes
s of
Pr.
30
0 m
and
11
not i
sola
ted.
1
Isol
atio
n: n
o ef
fect
on
abun
danc
e an
d
sam
plin
g pe
riod.
ric
hnes
s of
H; –
effe
ct o
n ric
hnes
s
of
Pr.
1 T
he c
itatio
n nu
mbe
rs c
orre
spon
d to
the
supe
rscr
ipt n
umbe
rs in
Tab
les
2, 3
, and
4.
2 O
bser
vatio
nal s
tudi
es a
re c
alle
d na
tura
l (se
e T
able
2)
if th
e ho
st p
lant
dis
trib
utio
n is
gen
eral
ly u
ndis
turb
ed b
y pe
ople
, or
anth
ropo
geni
c (s
ee T
able
3)
if th
e pl
ant d
istr
ibu-
tion
has
been
fund
amen
tally
cha
nged
by
hum
ans
(see
Tab
le 3
). In
the
expe
rimen
tal s
tudi
es (
see
Tab
le 4
) th
e co
nditi
ons
have
bee
n m
anip
ulat
ed b
y th
e re
sear
cher
(s).
3 In
sect
s ar
e id
entifi
ed a
s he
rbiv
ores
(H
); fu
ngiv
ores
(F
); tr
ap n
estin
g be
es (
B);
par
asito
ids
(P);
or
pred
ator
s (P
r). C
riter
ia e
valu
ated
are
dec
reas
e in
pat
ch s
ize
mea
sure
d as
are
a or
num
ber
of p
lant
s (p
atch
siz
e);
incr
ease
in is
olat
ion
mea
sure
d as
dis
tanc
e or
con
nect
ivity
(is
olat
ion
or f
ragm
enta
tion)
; de
crea
sing
yea
rs s
ince
fra
gmen
tatio
n (p
atch
age
); h
isto
ry o
f dis
turb
ance
(di
stur
banc
e); d
ista
nce
mov
ed b
y in
sect
s (d
ista
nce
mov
ed);
Typ
e of
mat
rix s
urro
undi
ng p
atch
es (
mat
rix).
The
effe
cts
of fr
agm
enta
tion
are
desi
gnat
ed a
s ne
gativ
e (–
) or
pos
itive
(+
).
ANN. ZOOL. FENNICI Vol. 42 • Fragmentation and trophic level 439
(such as Roland & Taylor 1997, Cappuccino et al. 1998, Komonen et al. 2000). Quality was generally density of host plant, but in some cases included plant size or the number of leaves or inflorescences. Several studies, such as Cronin and Haynes (2004) and Roland and Taylor (1997) also included aspects of the surroundings or the matrix. Once the data were collected they were presented as species occupancy or abundance, or species richness and diversity. When entire com-munities were involved, taxonomic groups were identified to species or lumped into functional groups or guilds.
Results
In all three types of studies of naturally (Table 2), anthropogenically (Table 3) and experimentally (Table 4) fragmented habitats, there were exam-ples of positive, neutral and negative effects of fragmentation. The negative effects on para-sitoids and predators were classified as “more negative” than for the herbivores if they declined with fragmentation but their hosts did not. Addi-tionally, in some cases the authors explicitly demonstrated that the negative effects for the higher trophic level were greater than would be expected simply due to decline in host/prey den-sity that went along with fragmentation.
In the 11 studies of naturally fragmented sys-tems (Table 2), there was little negative effect of habitat patch connectivity on herbivores. This is
not particularly surprising, because the insects in these systems are already known to persist in the fragmented landscape, though presumably there could still be effects of connectivity on patch occupancy. In five studies parasitoid abundance decreased with increasing patch isolation sug-gesting that at least some parasitoids in “natu-rally” fragmented habitats may be constrained by the spatial distribution of their hosts, but in seven studies there was no negative effect of connec-tivity. The effect of patch size and quality also varied among studies, but at least for herbivores, it had more of a negative effect than patch con-nectivity. The surrounding habitat or matrix was shown to be related to the distribution of parasi-toid in two very different ways. In one, Cappuc-cino et al. (1998) suggest that mixed deciduous forest surrounding isolated Balsam fir patches may enhance the presence of some Spruce bud-worm (Choristoneura fumiferana) parasitoids by providing alternate prey or adult food sources. Such spillover between habitats is discussed by Tscharntke et al. (2005). In contrast, Cronin and Haynes (2004) suggest that in tall grass prairie local extinction of both a herbivore and its para-sitoid is promoted by emigration from patches embedded in a smooth brome Bromus inermis matrix rather than in bare mudflats.
The negative effects of fragmentation on her-bivores, in terms of both reduced patch size and to a lesser extent connectivity, are perhaps more apparent in the eight studies of anthropogeni-cally fragmented systems (Table 3). The reduc-
Table 2. The effects of habitat fragmentation on herbivores and parasitoids in observational studies of naturally fragmented habitats.
Fragment Herbivoresa Parasitoids/predatorsa
positive none negative none/positive negative more negative
Size (decrease) 49, 110b, 11a 16, 8, 10a, 23, 26, 1+8, 189, 16, 26, y10a, 123
213 y10b,913 311a Connectivity (decrease) 12, 10a, b, 11a, 16, 8, 26, 29 1+2, 18, 15,179, 19, 11a, 913 16, 26
49, 213,1015 y10a,b, 211a
Age (–)/disturbance(+) y15 y15 115
Quality 49 y9
Matrixb 16, 8 1+6, 18
a Number of insect species or, if number is large but unspecified, then “y” indicates group effect. Superscript num-bers correspond to citation numbers in Table 1. A positive effect on parasitoids/predators is indicated by “+”.b See citation for explanations of matrix types. A single matrix type can be negative for some insects and positive for others.
440 van Nouhuys • ANN. ZOOL. FENNICI Vol. 42
tion of patch size had a larger effect on parasi-toids than on their hosts in three out of six cases, but the effect of connectivity on parasitoids was equivocal. There was little effect of the type of matrix on the herbivores. The matrix appeared to have a negative effect on several parasitoids, but a positive effect on one. Roland and Taylor (1997) attribute this to species-specific effects of surrounding forest (matrix) on connectivity.
The pattern is least clear in the collection of 11 experiments explicitly designed to test the effects of habitat fragmentation on herbiv-ores and their predators or parasitoids (Table 4). There is no consensus on the effects of patch size or connectivity. This result is similar to the review of 20 fragmentation experiments by Debinski and Holt (2000).
There are three possible explanations for
the variation of the effects of fragmentation in these studies. One is that the spatial scales differ greatly. At one extreme are the two studies of the insect communities on goldenrod (Kareiva 1987, Golden & Crist 1999) in which patches and distance between patches are on the scale of less than one to several meters. In these cases fragmentation is on a small scale relative to the movement of individual predators and parasi-toids and some herbivores. Hence, the mecha-nisms behind the effects of fragmentation relate to local foraging decisions and efficiency. In the case of Kareiva (1987), the herbivorous hosts are aphids that do not in fact move among patches during the experiment, but the predators do. At the other extreme is the study by J. A. Elzinga et al. (unpubl. data) in which patches were distrib-uted over 2 km. In these cases, the effects of hab-
Table 3. The effects of habitat fragmentation on herbivores and parasitoids in observational studies of anthropo-genically fragmented habitats.
Fragment Herbivoresa Parasitoids/predatorsa
positive none negative none/positive negative more negative
Size (decrease) 419a, 820, y28 322, 824 1019a, 1520,921
y21, 28, 224
Connectivity (decrease) 14, 820, y28 117, 22 4+4, 1520,1+22 14, 322,y17, 28
Age (–)/disturbance(+) 224 224 y17
Quality Matrixb 921 1+22 621 , 222
a Number of insects or, if number is large but unspecified, then “y” indicates group effect. Superscript numbers cor-respond to citation numbers in Table 1. A positive effect on parasitoids/predators is indicated by “+”.b See citation for explanations of matrix types.
Table 4. The effects of habitat fragmentation on herbivores and parasitoids in fragmentation experiments.
Fragment Herbivoresa Parasitoids/predatorsa
positive none negative none/positive negative more negative
Size (decrease) y3 15, 11b y3, 15, Connectivity (decrease) 214, 116 11, 312,y14 17, 212, y14, 25, 17, 211b, 1+12, 16, 21, 111b, 1218
818, 419b y12, 14 y12, 25, 419b Age (–)/disturbance(+) Quality 111b 111b
Matrixb 15 15
a Number of insect species or, if number is large but unspecified, then “y” indicates group effect. Superscript num-bers correspond to citation numbers in Table 1. A positive effect on parasitoids/predators is indicated by “+”.b See citation for explanations of matrix types. A single matrix type can be negative for some insects and positive for others.
ANN. ZOOL. FENNICI Vol. 42 • Fragmentation and trophic level 441
itat fragmentation may be explained by migra-tion rather than within-patch foraging behavior. These two scales are profoundly different, to a large extent because one is primarily explained by local interactions and the other is explained by landscape level spatial processes. Yet, the same language (fragmentation, patch, isolation …) is used to describe both. Presumably habitat fragmentation at all scales might influence spe-cies interactions in some way. What is important is for researchers to specify the scale they are working on relative to the landscape inhabited by the species, and ideally, the scale of move-ment of the species. Similarly, the temporal scale should be presented in terms of the generation times of the species involved (for a good exam-ple see Amarasekare 2000a, 2000b).
The second explanation for a lack of resolu-tion is that some of these studies (such as Kareiva 1987, Amarasekare 2000b) are about particular pre-chosen individual species, whereas many of the other studies are of entire communities (such as Golden & Crist 1999, Kruess & Tscharntke 2000b). The parasitoids and predators that are known well enough to be singled out a priori are unlikely to be a random subset of the species involved. Specifically, they may be more abun-dant and larger. Some of the strongest support for the idea that habitat fragmentation has a large effect on high trophic levels is in the form of species richness data or total parasitism, and sev-eral studies made the important point that habitat fragmentation had the largest effect on species that were rare in the first place (Golden & Crist 1999, Kruess & Tscharntke 2000b). In contrast, some of the most convincing demonstrations of a lack of fragmentation effects emerge from stud-ies of individual species — even when they were studied on a relatively large spatial scale (such as van Nouhuys & Hanski 2002, discussed below).
A third reason that the fragmentation experi-ments may present different results than the observational studies of naturally or anthropon-genically fragmented settings is that the specific aspect being tested is the dispersal ability of the target species. The distance moved by individu-als, and their propensity to move, is probably the main factor explaining their pattern of habitat use in the fragmentation experiments. This is because the experiments lasted only a short
time. This short time frame may predispose the parasitoids to be more affected than their prey, because in order for the parasitoids to colonize a fragment, the host must have already colonized it. In fact, none of the fragmentation experiments explored the effects of fragmentation over time. Experimental studies that address both dispersal and persistence may be quite fruitful. Several observational studies that measured colonization (such as Cappuccino et al. 1998, van Nouhuys & Hanski 2000) got around the bias against coloni-zations by parasitoids relative to herbivores by assessing the effects of fragmentation separately for the herbivore and the parasitoid.
Case study
We gain further insight into the complexity of the relationship of trophic level with habitat frag-mentation by looking in detail at one research system. The parasitoid community associated with the Glanville fritillary Melitaea cinxia (Lepidoptera: Nymphalidae) lends itself well to this task because there are only a few species involved, and these species cover four trophic levels and a range of different life histories. In addition, a long term study of this group of interacting species at both small and large spatial scales has resulted in the development of mecha-nistic explanations for the observed patterns of habitat use. I will briefly introduce the species involved, and then explain how I think habitat fragmentation affects each species separately, and the interactions among them.
The landscape, the host plants and the butterfly
The Glanville fritillary butterfly inhabits steppe and open meadow habitats in Europe and Asia (Wahlberg et al. 2004). In Finland, M. cinxia is restricted to the Åland Islands in the Baltic Sea, where it feeds on Plantago lanceolata and Veronica spicata. The Åland Islands are separated from other M. cinxia populations in Sweden and Estonia by several hundred km. The butterfly persists as a classical metapopulation comprised of 300 to 500 local populations with a
442 van Nouhuys • ANN. ZOOL. FENNICI Vol. 42
high rate of turnover (local extinctions and colo-nizations) (Hanski et al. 1995, Ehrlich & Hanski 2004, Hanski & Meyke 2005). There are nearly 4000 suitable habitat patches for local popula-tions in an area of 50 ¥ 70 km (Fig. 1). Habitat patches range in size from less than 100 m2 to 10 hectares (average less than 0.5 hectare). Most butterflies stay within their natal patch to breed. Local butterfly populations range in size from one to tens of gregarious larval nests, and are ephemeral, most persisting only a few years. About 25% have persisted the entire time they have been surveyed (14 years). The regional landscape is structured so that patches are clus-tered into 130 semi-independent patch networks separated by small-scale agriculture, water and forest. Mark-release-recapture studies and survey data show that there is very little movement of butterflies between patch networks, but signifi-cant movement among local populations within them (Nieminen et al. 2004).
The parasitoids
The natural enemy community associated with M. cinxia in Åland is relatively simple. There are two primary larval parasitoids that use only M. cinxia as a host, and each of these parasitoids has a closely associated hyperparasitoid. Less inte-gral species in the community are several gen-
eralist pupal parasitoids (Lei et al. 1997), coc-cinelid beetle and lacewing (chrysopid) larvae that feed opportunistically on eggs and small larvae, predatory Hemiptera sometimes feeding on larvae, and spiders that occasionally capture adult butterflies (van Nouhuys & Hanski 2004).
One of the primary larval parasitoids is Cotesia melitaearum (Wilkinson) (Ichneumo-noidae: Braconidae). This wasp is a parasitoid of several species of checkerspot butterflies in Europe and Asia. Cotesia melitaearum is a gre-garious endoparasitoid, laying one to about 40 eggs inside a host larva, depending on the size of the host. There are two or sometimes three generations per year and host generation, and C. melitaearum spends the winter as a larva inside the host larva (van Nouhuys & Lei 2004). In the Åland Islands C. melitaearum has a metap-opulation structure, and is extremely sensitive to habitat fragmentation and the turnover (extinc-tions and colonizations) of local host (M. cinxia) populations. Currently there are few small local populations (Fig. 1) (Lei & Hanski 1997, van Nouhuys & Hanski 2002).
Cotesia melitaearum is parasitized by Gelis agilis (Fabricius) (Ichneumonidae: Cryptinae). Female wasps in the genus Gelis are wingless generalist ectoparasitoids, and males are rare or unknown. In the Åland Islands Gelis agilis and several related species are hyperparasitoids of C. melitaearum pupae. Gelis have other hosts
Fig. 1. Map of the Åland Islands showing locations of habitat patches unoc-cupied by the host butter-fly, Melitaea cinxia (small white circles), patches occupied by M. cinxia and the parasitoid Hyposoter horticola (small black cir-cles), and by the parasi-toid Cotesia melitaearum (large white circles in the west) in 2004. The hyper-parasitoid Mesochorus sp. cf. stigmaticus is absent from the islands enclosed by the square in the east.
ANN. ZOOL. FENNICI Vol. 42 • Fragmentation and trophic level 443
in Åland (though the species are not known), and can be extremely abundant, aggregating where there is a high density of C. melitaearum cocoons, and having a dramatic density depend-ent effect on C. melitaearum, even causing local extinctions (Lei & Hanski 1998, van Nouhuys & Hanski 2000, van Nouhuys & Tay 2001).
The second primary parasitoid of the butterfly M. cinxia is Hyposoter horticola (Gravenhorst) (Ichneumonidae: Campoplaginae). This wasp is also a parasitoid of checkerspot butterflies in Europe and Asia, but its only confirmed host is M. cinxia (van Nouhuys & Hanski 2004). It is a solitary endoparasitoid, laying eggs in first instar host larvae just before the larvae hatch from the egg (van Nouhuys & Ehrnsten 2004). In the Åland Islands there is one generation per year. Hyposoter horticola parasitizes about a third of the larvae in virtually all of the local host popu-lations of M. cinxia in Åland (Fig. 1). The wasp maintains this remarkably uniform rate of para-sitism because it is more dispersive than the host, and because it forages efficiently by learning the locations of multiple host egg clusters before the hosts are ready to be parasitized (van Nouhuys & Ehrnsten 2004, van Nouhuys & Hanski 2002).
The secondary parasitoid of H. horticola is Mesochorus sp. cf. stigmaticus (Brischke) (Ichneumonidae: Mesochorinae). Wasps in the genus Mesochorus are generally endoparasitic hyperparasitoids. The taxonomy in the genus is not well worked out, but it appears that Meso-chorus sp. cf. stigmaticus is associated only with parasitoids of checkerspot butterflies (Lei et al. 1997). In the Åland Islands it is associated with M. cinxia, hyperparasitizing about a quarter of the H. horticola, and occasionally C. melitae-arum as well.
Fragmentation and the second and third trophic levels
More than a decade of survey data, modeling and several large-scale mark-release-recapture studies show that the distribution of the butterfly, M. cinxia, is constrained by the fragmentation of habitat available in Åland (Nieminen et al. 2004, Hanski & Meyke 2005). One demonstration of this is the pattern of occupancy of habitat patch
networks of differing configuration. Each of the 130 semi-independent patch networks in Åland differ in size, number and connectivity of suita-ble habitat patches. The “metapopulation capac-ity” is an index of the ability of a patch network to support a metapopulation of the butterfly, based on habitat area and connectivity (Hanski & Ovaskainen 2000). Only those networks with a high metapopulation capacity are occupied by butterflies.
The two primary parasitoids use the same resource (host) in the same habitat, but differ greatly in their abundance, distribution (Fig. 1) and dynamics. A large part of this difference is explained by their differing abilities to use a host in a fragmented habitat. Hyposoter horticola is not affected by the present fragmentation of the habitat. It experiences a single, perhaps patch-ily distributed, host population. The abundance and rate of parasitism is not affected by the size or isolation of local host populations and the wasp is able to colonize new host populations the same year as they are themselves initiated. Cotesia melitaearum, on the other hand, has a metapopulation structure. Based on annual survey and dispersal experiments it is evident that the wasp is more constrained by habitat fragmentation than the host is. This is probably due to the range (distance traveled) rather than rate of dispersal (propensity to move) of C. meli-taearum (van Nouhuys & Hanski 2002).
Movement within the landscape is not the only difference between the two wasps that is related to their sensitivity to habitat fragmenta-tion. Cotesia melitaearum is gregarious and has several generations per year. This means that it has a much higher potential rate of increase than the host, or H. horticola, which is solitary and has only one generation per year. Increased population size does help mitigate the effects of fragmentation for C. melitaearum, but large population size is rarely realized, and is damp-ened by density dependent hyperparasitism (van Nouhuys & Tay 2001, van Nouhuys & Lei 2004).
One might presume that because the two par-asitoids differ in dispersal behavior their coexist-ence would be facilitated by a trade off between dispersal ability and local competitive ability (Lei & Hanski 1998, Holt 2002). However, it
444 van Nouhuys • ANN. ZOOL. FENNICI Vol. 42
turns out that H. horticola is superior both as a disperser and as a local competitor. Hence, in this case, the inferior disperser C. melitaearum persists by using host left unused by H. horti-cola, and because of its high rate of reproduction (van Nouhuys & Hanski 2004).
Fragmentation and the fourth trophic level
Fragmentation of the habitat suitable for the butterfly probably has little impact on the hyper-parasitoid Gelis agilis. This is because it has host species other than C. melitaearum (Schwarz & Shaw 1999), that presumably occupy other sur-rounding habitats. Indirect evidence of this is the fast rate that individuals are recruited, on foot (because they are wingless), to concentrations of C. melitaearum cocoons. Mesochorus sp. cf. stigmaticus, on the other hand, is restricted to the parasitoids of M. cinxia, and to H. hor-ticola in particular. One would expect that a fourth trophic level specialist parasitoid would be constrained by habitat fragmentation. How-ever, M. stigmaticus appears to be quite abun-dant throughout most of Åland. This is probably because H. horticola is a reliable resource even though it is at the third trophic level, and because M. stigmaticus is dispersive, like H. horticola. The one part of Åland where M. stigmaticus is absent is two small patch networks on isolated islands in the east (Fig. 1). Perhaps the absence of M. stigmaticus from these isolated poor qual-
ity networks can be attributed to them being too small to support four trophic levels (Holt 2002).
Conclusion
The case study of the parasitoids associated with the Glanville fritillary butterfly provides a clear illustration of the roles of dispersal behavior and host range in defining the range of effects of fragmentation for third and fourth trophic level species (Fig. 2). Other characteristics of species, such as their longevity and reproductive capacity, must also contribute to their sensitivity to fragmentation (Henle et al. 2004). Unfortu-nately, the natural history of higher trophic level insect species are rarely known well enough for specific predictions to be made for individual parasitoids or predators.
The results of individual fragmentation stud-ies differed greatly. Some of this variation can be attributed to variation in definitions of frag-mentation, durations of studies, and types of data collected. However, some is also due to interspe-cific differences in the effects of fragmentation, independent of trophic level, as was illustrated in the case study. With this much variation among species, the effect of fragmentation on different trophic levels may be detected most easily by gathering data about an entire community and recording species richness and diversity, and the distribution of rare vs. abundant species. These studies of entire communities show the strongest effects. However, detailed studies of individual species are necessary to understand many of the characteristics that make species sensitive to fragmentation. It may well be that on aver-age, fragmentation affects the third trophic level (parasitoids and predators) more than the second trophic level (herbivorous insects), but this is not supported by Tables 2, 3 and 4.
An important implication for conservation from the case study is that fragmentation leads toward the loss of sedentary and specialized species at higher trophic levels. In fact, the para-sitoid C. melitaearum (Figs. 1 and 2) is likely to become extinct in the Åland Islands (van Nouhuys & Hanski 2002). Based on several of the studies in the literature review, rarity should be on the list of attributes of fragile species, in
Fig. 2. Schematic drawing of the relationship between habitat fragmentation and the persistence of popula-tions for the four parasitoids in the Glanville fritillary butterfly system. (a) The hyperparasitoid Gelis agilis, (b) the hyperparasitoid Mesochorus sp. cf. stigmaticus, (c) the parasitoid Hyposoter horticola, and (d) the para-sitoid Cotesia melitaearum.
ANN. ZOOL. FENNICI Vol. 42 • Fragmentation and trophic level 445
addition to sedentariness and specialization. That is, parasitoids or predators that are initially rare in the landscape rapidly become increasingly rare with fragmentation (Dubbert et al. 1998, Golden & Crist 1999, Komonen et al. 2000, Kruess & Tscharntke 2000a, 2000b). The next important question is what happens to communi-ties with the loss of sedentary, specialized, and rare higher trophic level species (for related dis-cussion see Morris et al. 2005). If sedentary spe-cialized parasitoids generally have strong effects on local host dynamics, as C. melitaearum once had in Åland (Lei & Hanski 1997), then commu-nity structure may change greatly. Thus, while the overall effects of habitat fragmentation may not be stronger for higher trophic levels than for the species upon which they depend, specific key high level community members may be affected in ways that disrupt community structure.
Acknowledgements
This would not have been written if T. Roslin and J. Kotze had not invited me to participate in the workshop on spatial ecology of insect–plant interactions, so I thank them and the other participants in the workshop. I also thank T. Tscharn-tke, T. Roslin, J. Kotze and an anonymous reviewer for their helpful comments on the manuscript.
References
Amarasekare, P. 2000a: Coexistence of competing parasi-toids on a patchily distributed host: Local vs. spatial mechanisms. — Ecology 81: 1286–1296.
Amarasekare, P. 2000b: Spatial dynamics in a host-multi-parasitoid community. — J. Anim. Ecol. 81: 201–213.
Antolin, M. F. & Strong, D. R. 1987: Long distance dispersal by a parasitoid (Anagrus delicatus, Mymaridae) and its host. — Oecologia 73: 288–292.
Askew, R. R. & Shaw, M. R. 1989: Parasitoid communities: their size, structure and development. — In: Waage, J. & Greathead, D. (eds.), Insect parasitoids: 225–264. Aca-demic Press, San Diego.
Athen, O. & Tscharntke, T. 1999: Insect communities of Phragmites habitats used for sewage purification: effects of age and area of habitats on species richness and herbivore–parasitoid interactions. — Limnologica 29: 71–74.
Braschler, B., Lampel, G. & Baur, B. 2003: Experimental small-scale grassland fragmentation alters aphid popula-tion dynamics. — Oikos 100: 581–591.
Brodeur, J. 2000: Host specificity and trophic relationships of
hyperparasitoids. — In: Hochberg, M. E. & Ives, A. R. (eds.), Parasitoid population biology: 163–183. Prince-ton University Press, San Diego.
Cappuccino, N., Lavertu, D., Bergeron, Y. & Regniere, J. 1998: Spruce budworm impact, abundance and para-sitism rate in a patchy landscape. — Oecologia 114: 236–242.
Cronin, J. T. 2003: Patch structure, oviposition behavior, and the distribution of parasitism risk. — Ecol. Monogr. 73: 283–300.
Cronin, J. T. 2004: Host–parasitoid extinction and coloniza-tion in a fragmented prairie landscape. — Oecologia 139: 503–514.
Cronin, J. T. & Haynes, K. J. 2004: An invasive plant pro-motes unstable host–parasitoid patch dynamics. — Ecol-ogy 85: 2772–2782.
Cronin, J. T., Haynes, K. J. & Dillemuth, F. 2004: Spider effects on planthopper mortality, dispersal, and spatial population dynamics. — Ecology 85: 2134–2143.
Cronin, J. T., Reeve, D. J., Wilkens, R. & Turchin, P. 2000: The pattern and range of movement of a checkered beetle predator relative to its bark beetle prey. — Oikos 90: 127–138.
Debinski, D. M. & Holt, R. D. 2000: A survey and overview of habitat fragmentation experiments. — Conserv. Biol. 14: 1–13.
Didham, R. K., Ghazoul, J., Stork, N. E. & Davis, A. J. 1996: Insects in fragmented forests: a functional approach. — Trends Ecol. Evol. 11: 255–260.
Dubbert, M., Tscharntke, T. & Vidal, S. 1998: Stem-boring insects of fragmented Calamagrostis habitats: herbiv-ore–parasitoid community structure and the unpredict-ability of grass shoot abundance. — Ecol. Entomol. 23: 271–280.
Eber, S. 2001: Multitrophic interactions: The population dynamics of spatially structured plant–herbivore–parasi-toid systems. — Basic Appl. Ecol. 2: 27–33.
Ehrlich, P. R. & Hanski, I. (eds.) 2004: On the wings of checkerspots: A model system for population biology. — Oxford University Press, Oxford.
Faeth, S. H. & Simberloff, D. 1981. Experimental isolation of oak host plants: effects on mortality, survivorship, and abundances of leaf-mining insects. — Ecology 63: 625–635.
Ferrari, J., Kruess, A. & Tscharntke, T. 1997: Effects of fragmentation of wild roses on their insect communities. — Mitteilungen der Deutschen Gesellschaft fur Allge-meine und Angewandte Entomologie 11: 89–90.
Golden, D. M. & Crist, T. O. 1999: Experimental effects of habitat fragmentation on old-field canopy insects: com-munity, guild and species responses. — Oecologia 118: 371–380.
Hanski, I. & Meyke, E. 2005: Large-scale dynamics of the Glanville fritillary butterfly: landscape structure, popula-tion processes, and weather. — Ann. Zool. Fennici 42: 379–395.
Hanski, I. & Ovaskainen, O. 2000: The metapopulation capacity of a fragmented landscape. — Nature 404: 755–758.
Hanski, I., Pakkala, T., Kuussaari, M. & Lei, G. C. 1995:
446 van Nouhuys • ANN. ZOOL. FENNICI Vol. 42
Metapopulation persistence of an endangered butterfly in a fragmented landscape. — Oikos 72: 21–28.
Harrison, S. & Thomas, C. D. 1991: Patchiness and spatial pattern in the insect community on ragwort Senecio jacobaea. — Oikos 62: 5–12.
Hassell, M. P. 2000: The spatial and temporal dynamics of host-parasitoid interactions. — Oxford University Press, London.
Henle, K., Davies, K., Kleyer, M., Margules, C. & Settele, J. 2004: Predictors of species sensitivity to fragmentation. — Biodivers. Conserv. 13: 207–251.
Holt, R. D. 2002: Food webs in space: On the interplay of dynamic instability and spatial processes. — Ecol. Res. 17: 261–273.
Kareiva, P. 1987: Habitat fragmentation and the stability of predator-prey interactions. — Nature 326: 388–390.
Komonen, A., Penttilä, R., Lindgren, M. & Hanski, I. 2000: Forest fragmentation truncates a food chain based on an old-growth forest bracket fungus. — Oikos 90: 119–126.
Kruess, A. 2003: Effects of landscape structure and habi-tat type on a plant–herbivore–parasitoid community. — Ecography 26: 283–290.
Kruess, A. & Tscharntke, T. 1994: Habitat fragmentation, species loss and biological control. — Science 264: 1581–1584.
Kruess, A. & Tscharntke, T. 2000a: Effects of habitat frag-mentation on plant-insect communities. — In: Ekbom, B., Irwin, M. E. & Roberts, Y. (eds.), Interchanges of insects between agricultural and surrounding land-scapes: 53–70. Kluwer, London.
Kruess, A. & Tscharntke, T. 2000b: Species richness and parasitism in a fragmented landscape: experiments and field studies with insects on Vicia sepium. — Oecologia 122: 129–137.
Lei, G. C. & Hanski, I. 1997: Metapopulation structure of Cotesia melitaearum, a specialist parasitoid of the but-terfly Melitaea cinxia. — Oikos 78: 91–100.
Lei, G. C. & Hanski, I. 1998: Spatial dynamics of two com-peting specialist parasitoids in a host metapopulation. — J. Anim. Ecol. 67: 422–433.
Lei, G. C., Vikberg, V., Nieminen, M. & Kuussaari, M. 1997: The parasitoid complex attacking the Finnish populations of Glanville fritillary Melitaea cinxia (Lep: Nymphalidae), an endangered butterfly. — J. Nat. Hist. 31: 635–648.
MacArthur, R. H. & Levins, R. 1964: Competition, habitat selection, and character displacement in a patchy environ-ment. — Proc. Natl. Acad. Sci. U.S.A. 51: 1207–1210.
Mikkelson, G. M. 1993: How do food webs fall apart? A study of changes in trophic structure during relaxation on habitat fragments. — Oikos 67: 539–547.
Moilanen, A. & Nieminen, M. 2002: Simple connectivity measures for spatial ecology. — Ecology 84: 1131–1145.
Morin, P. J. 1999: Community ecology. — Blackwell Science Inc., Malden.
Morris, R. J., Lewis, O. T. & Godfray, H. C. J. 2005: Apparent competition and insect community structure: towards a spatial perspective. — Ann. Zool. Fennici 42: 449–462.
Nieminen, M., Siljander, M. & Hanski, I. 2004: Structure and dynamics of Melitaea cinxia metapopulations. — In: Ehrlich, P. R. & Hanski, I. (eds.), On the wings of check-erspots: a model system for population biology: 63–91. Oxford University Press, Oxford.
Pimm, S. L. 1982: Food webs. — Chapman and Hall, London.
Roland, J. & Taylor, P. D. 1997: Insect parasitoid species respond to forest structure at different spatial scales. — Nature 386: 710–713.
Schwarz, M. & Shaw, M. R. 1999: Western Palaearctic Cryptinae (Hymenoptera: Ichneumonidae) in the National Museums of Scotland, with nomenclatural changes, taxonomic notes, rearing records and special reference to the British check list. Part 2. Genus Gelis Thunberg (Phygadeuontini: Gelina). — Entomologist’s Gaz. 50: 117–125.
Segarra-Carmona, A. E. & Barbosa, P. 1992: Host-plant patches as islands: effects of patch size, patch extinc-tion and seasonality of a herbaceous tropical legume (Crotalaria pallida Ait.) on a pod borer (Etiella zinck-enella (Treit.) and its parasitoids. — Insect Sci. Appl. 13: 709–718.
Shaw, M. R. 1994: Parasitoid host ranges. — In: Hawkins, B. A. & Sheehan, W. (eds.), Parasitoid community ecology: 111–144. Oxford University Press, Oxford.
Tscharntke, T. & Kruess, A. 1999: Habitat fragmentation and biological control. — In: Hawkins, B. A. & Cornell, H. V. (eds.), Theoretical approaches to biological control: 190–205. Cambridge University Press, Cambridge.
Tscharntke, T., Gathmann, A. & Steffandewenter, I. 1998: Bioindication using trap-nesting bees and wasps and their natural enemies: Community structure and interac-tions. — J. Anim. Ecol. 35: 708–719.
Tscharntke, T., Rand, T. A. & Bianchi, F. J. J. A. 2005: The landscape context of trophic interactions: insect spillover across the crop–noncrop interface. — Ann. Zool. Fennici 42: 421–432.
van der Meijden, E. & van der Veen-van Wijk, C. A. M. 1997: Tritrophic metapopulation dynamics: a case study of ragwort, the Cinnabar moth and the parasitoid Cotesia popularis. — In: Hanski, I. & Gilpin, M. E. (eds.), Meta-population biology: ecology, genetics, and evolution: 387–406. Academic Press, San Diego.
van Nouhuys, S. & Ehrnsten, J. 2004: Wasp behavior that leads to uniform parasitism of a host available only a few hours per year. — Behav. Ecol. 15: 661–665.
van Nouhuys, S. & Hanski, I. 1999: Host diet affects extinc-tions and colonizations in a parasitoid metapopulation. — J. Anim. Ecol. 68: 1248–1258.
van Nouhuys, S. & Hanski, I. 2000: Apparent competition between parasitoids mediated by a shared hyperparasi-toid. — Ecol. Lett. 3: 82–84.
van Nouhuys, S. & Hanski, I. 2002: Colonisation rates and distances of a host butterfly and two specific parasi-toids in a fragmented landscape. — J. Anim. Ecol. 71: 638–650.
van Nouhuys, S. & Hanski, I. 2004: Natural enemies of checkerspot butterflies. — In: Ehrlich, P. R. & Hanski, I. (eds.), On the wings of checkerspots: a model system for
ANN. ZOOL. FENNICI Vol. 42 • Fragmentation and trophic level 447
population biology: 161–180. Oxford University Press, Oxford.
van Nouhuys, S. & Hanski, I. 2005: Metacommunities of butterflies, their host plants and their parasitoids. — In: Holyoak, M., Leibold, M. A. & Holt, R. D. (eds.), Meta-communities: spatial dynamics and ecological communi-ties. University of Chicago Press, Chicago. [In press].
van Nouhuys, S. & Lei, G. C. 2004: Parasitoid and host metapopulation dynamics: The influences of temperature mediated phenological asynchrony. — J. Anim. Ecol. 73: 526–535.
van Nouhuys, S. & Tay, W. T. 2001: Causes and conse-
quences of mortality in small populations of a parasitoid wasp in a fragmented landscape. — Oecologia 128: 126–133.
Wahlberg, N., Ehrlich, P. R., Boggs, C. L. & Hanski, I. 2004: Bay checkerspot and Glanville fritillary compared with other species. — In: Ehrlich, P. R. & Hanski, I. (eds.), On the wings of checkerspots: a model system for population biology: 219–244. Oxford University Press, Oxford.
Zabel, J. & Tscharntke, T. 1998: Does fragmentation of Urtica habitats affect phytophagous and predatory insects differently? — Oecologia 116: 419–425.
This article is also available in pdf format at http://www.sekj.org/AnnZool.html